supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
<pie_>
supercuts? <gchristensen> pie_: yeah we tried to make supercuts but it became annoying
cole-h has joined #nixos-chat
<samueldr>
pie_: imagine an [apple]CEO's keynote reduced to all the times they say "magic"
<pie_>
screenshiot? <tazjin> drakonis: fwiw, the initial implementation (which doesn't do everything in the screenshot) is here: https://cl.tvl.fyi/c/depot/+/2237
<drakonis>
yeah i saw it
<drakonis>
its a tweet i linked earlier
<pie_>
drakonis: no i mean what was in the screenshot
<pie_>
was it the lisp transform? i did fidn that meanwhile
<eyJhb>
ashkitten: isn't it because that many features that are requested in wlroots, is by the sway devs? Also seem to think, that they are actually the same developers
<pie_>
ill look at that but its still bs that i need to do this
<pie_>
"Decrypt Signal encrypted backups outside the app " well that will basically do what i want but id really jsut want my damn messages copied to the desktop version...
<pie_>
i should be studying for my test tomorrow instead of dealing with this crap
* pie_
sighs and adds "unf*** signal" to todo list
<__monty__>
Or convince people to switch to a service that makes fewer assumptions on behalf of its users : )
<lovesegfault>
gchristensen: can I ask you for help testing/understanding something?
<gchristensen>
you can try :)
<lovesegfault>
Alright, let's check my assumptions first: if I have foo = { ... }: { }, then I can move { ... }: { } into foo.nix and say `foo = import ./foo.nix`
<gchristensen>
yes, assuming foo's definition doesn't use anything in a let or rec expression
<lovesegfault>
so I should be able to do that here
<lovesegfault>
just say `outputs = import ./output.nix`
<lovesegfault>
;
<gchristensen>
not sure... flakes might add some more rules
<lovesegfault>
:(
<gchristensen>
but I would expect that to work if it were called foo.nix
<lovesegfault>
Right, I know this works in general, but with flakes I get: expected a function but got a thunk at /nix/store/h21gfgrkjkhwl0f1ri2jhaqsdssrzm86-source/flake.nix:29:3
<__monty__>
Sounds like a laziness issue. Flakes not forcing the thunk so it can see it *is* a function.
<lovesegfault>
Right, that sounds like a bug?
<gchristensen>
it might be, worth reporting
<lovesegfault>
where to? nixos/nix?
<gchristensen>
yea
<lovesegfault>
coolio, on it
<lovesegfault>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 392
<lovesegfault>
__monty__++
<{^_^}>
__monty__'s karma got increased to 26
<pie_>
i need to break things more often so i run into issues sooner
<gchristensen>
I'll trade you
<patagonicus>
Why do browsers have to be so`complicated? On i3/X11 HiDPI just doesn't work with multiple screens and at least for a while Chrome was completely broken (bug only says that --disable-gpu fixes it, but nothing much has happened in the 2 months it was open) and with sway/Wayland it's blurry as hell on HiDPI screens and on my other machine Youtube
<patagonicus>
becomes really choppy sometimes and I have to restart Chrome (seems to be some combination of screen locking and/or having Chrome run without doing anything - having Youtube run for a few hours without a break is fine).
<lovesegfault>
I ended up moving to wayland to get that sort of stuff working
<patagonicus>
lovesegfault: Oh, yeah, forgot to mention that one my laptop Chromium with Wayland backend can't play Youtube and on my desktop it just doesn't even start.
<lovesegfault>
That's... new!
<lovesegfault>
I don't use chromium either, so idk :P
<patagonicus>
Yeah, I could never get used to Firefox and I don't think Opera has been a serious browser for a while now. And I don't even want to get started trying out other browsers.
<lovesegfault>
fwiw, chromium & chrome run well here
<lovesegfault>
just tested
<lovesegfault>
I'm on nixos-unstable-small though
<gchristensen>
chromium / chrome run in xwayland and xwayland doesn't support hi(?)mixed(?) dpi and looks fuzzy to me
<patagonicus>
gchristensen: chromium does have a native Wayland backend, but it's … experimental.
<patagonicus>
lovesegfault: It seems to heavily depend on your graphics card and graphics card driver. But you're right, I should try unstable. I think I went back before I switched to sway because Chrome was broken with my Intel GPU on unstable and X11.
<gchristensen>
how do I use it?
<lovesegfault>
gchristensen: enable it in the pkg
<lovesegfault>
Oh
<lovesegfault>
wait
<lovesegfault>
maybe it got removed?
<patagonicus>
No, at least Beta should have it. One sec.
<lovesegfault>
the package had a `enableOzone ? false` arg
<lovesegfault>
but idk where it went
<patagonicus>
I think that changed because it's now enabled by default. But, yeah, when I looked into it I also got really confused by the nixpkgs. It does change behavior and on my laptop it's no longer the blurry/scaling mess it is with XWayland, so it seems to work.
<{^_^}>
nix#4384 (by lovesegfault, 17 seconds ago, open): Can't eta-reduce with flakes: expected a function but got a thunk
<lovesegfault>
boom
<FRidh>
nix seems quite picky when it comes to flake.nix. Overlay has to be `final: prev:` for example
<lovesegfault>
Having moved to flakes this week the whole thing seems to be in a picky state :P
<gchristensen>
unreleased and experimental even :P
<FRidh>
still worth it imo
<patagonicus>
Well, on unstable sway fails to start, so I can't really test if Chrome would be doing better.
<gchristensen>
I've not used flakes yet. I should give it a go one of these days :')
<lovesegfault>
patagonicus: fails how?
<lovesegfault>
I'm running sway on unstable _right now_
* etu
recently migrated away from flakes
<lovesegfault>
gchristensen: I migrated my whole system config this week
<lovesegfault>
It was a bit hard to grok, but overall was ok
<lovesegfault>
found three bugs in deploy-rs in the process :P
<patagonicus>
Fails to create a GBM device, fails to intialize renderer, fails to open DRM device 9, fails to open any DRM device, unable to create backend in that order. Before that MESA-LOADER complains about not finding iris, kme_swrast or swrast.
<patagonicus>
lovesegfault: you running it on the same GPU I am? I said it was GPU and driver specific stuff. :P
<lovesegfault>
I run sway on my intel iGPU, on an AMD gpu, and on a RPI4
<FRidh>
etu: how so?
<lovesegfault>
it works on all of the above :P
<lovesegfault>
oh, and nvidia with nouveau
<patagonicus>
lovesegfault: want to give me your config for the Intel one? Must be doing something wrong here.
<patagonicus>
Guess I could try removing most of it.
<patagonicus>
Are the env variables anything that affects sway or are they just for stuff running under sway?
<lovesegfault>
the latter
<patagonicus>
Then I guess the only difference that would make sense to affect stuff is that in intel.nix you have libva-full (and intel-compute-engine) and I have vaapiIntel (and we both have intel-media-driver). Your sway config is just enabling sway, installing some tooling and setting the env stuff.
<lovesegfault>
uhhh, what's the flake-equivalent of nix-instantiate O.o
<lovesegfault>
patagonicus: I would be blown away if libva was the difference
<lovesegfault>
you have opengl enabled, right>
<lovesegfault>
*?
<patagonicus>
Yeah
<etu>
FRidh: Not stable feature, quite lacking docs since there's no official docs since it's not a stable feature. Then i encountered a thing that forces you to go all in on all your systems on flakes, not as nice to go for it on only one system...
<patagonicus>
Let me build a minimal configuration.nix to see if that works.
<patagonicus>
Err. Wait, maybe I managed to get it removed when I refactored my config. But then how was it working on stable? Let me try that.
<patagonicus>
No, no difference.
<patagonicus>
As in, nixos-rebuild didn't pull anything new in.
<FRidh>
yes with a single system for yourself its not really worth it. On my own systems I noticed getting updates is annoying; you effectively need to let CI do that. I think the autoUpgrade may need to be extended to make it more obvious how to get updates if you cannot fetch an external flake
<etu>
FRidh: But with multiple systems, you have to run all of them on flakes or have separate repos.
<etu>
FRidh: touch /etc/nixos/flake.nix breaks nixos rebuild on a non flake system
<FRidh>
etu: flake.nix does not have to be in the top-level of a repo
<FRidh>
although that is typically done
<etu>
FRidh: It's still broken that an empty file like that triggers a feature i haven't enabled to break because that feature isn't enabled.
<patagonicus>
lovesegfault: I just tried with this somewhat minimized config (still some stuff in there for / on LUKS and stuff like that): http://sprunge.us/7MZNV6 If you can find something that's wrong or missing, let me know. For now I'm switching back to stable.
<lovesegfault>
patagonicus: can you post the error too?
<bbigras>
lovesegfault: I wonder if the deploy-rs bug about setting the default boot entry or whatever is the same I had. When I switched to flakes I still had old boot entries. and when I booted up, systemd-boot selected the last of the old boot entries. I thought maybe it was because they were not named the same way. Deleting old entries fixed it for me.
<lovesegfault>
Yep, that's probably why bbigras
<lovesegfault>
I had that happen on all my systems
dadada_ has joined #nixos-chat
<lovesegfault>
well, all my systemd-boot ones
<bbigras>
I should have reported it
<bbigras>
luckily I didn't have the "activate" is still running forever bug.
<bbigras>
I'll try to reproduce the nix-build-uncached "bug" I have with github actions soon.
<bbigras>
lovesegfault: Would `nix flake check` build all the hosts before that change? I'm not sure if it's the same.
<lovesegfault>
I want the outputs so I can easily inspect
<bbigras>
gotcha. thanks. I'll try it.
cosimone has joined #nixos-chat
Fare has joined #nixos-chat
<gchristensen>
playing the "is this covid or is this a big coincidence?" game today
<abathur>
ruhroh
<gchristensen>
sore muscles (did demolition and moved furniture and rugs) sore throat (runny nose from allergens) tired (up late and often for last night)
<abathur>
yeah, not a fan of this game
<gchristensen>
not great!
<abathur>
I'm just glad taste randomly disappearing isn't just one of those frequent adult human experiences no one warns you about
<abathur>
that'd really turn it all up to 11
<gchristensen>
oh god
<gchristensen>
yeah I can taste
<lovesegfault>
does anyone know if you can do fancy stuff with gh actions? like only moving on to another set of jobs if the first set succees
<bbigras>
since the start of the pandemic it's been like that for me too. every single tiny symptoms made me doubt. I, too, have some allergies symptoms. I wonder if it's because of the cold outside or the heating inside.
<lovesegfault>
like, here I want `check-flake` to only happen if the rest succeeded
<lovesegfault>
I think I'm asking for a CI pipeline
<bbigras>
could you use the matrix thing with a condition? maybe conditions are more like "this build is for the release or not" and not based on the previous run.
<lovesegfault>
can I expand my whole matrix into a list?
<bbigras>
no idea. I'm just googling stuff. I didn't used matrix yet.
<lovesegfault>
:D
<bbigras>
can a matrix be used in a step?
<bbigras>
like build step would have a matrix will all the system and the next step would depend on the build step
<lovesegfault>
I'm trying to learn how to use pre-commit
<bbigras>
👍️
<sphalerite>
what's the preferred solution for network mounts on linux nowadays? I heard samba is the weapon of choice for both macos and windows, but I'm still uncertain about linux…
<lovesegfault>
sphalerite: nfs?
<bbigras>
does nfs still lock up if the network drops? or that is something that was decades ago?
<sphalerite>
Is there a good way to do auth with nfs nowadays?
Dotz0cat has quit [Read error: Connection reset by peer]
<bbigras>
lovesegfault: did you set the fast connection thing in outputs.nix? I need to do it but only for 1 host.
<lovesegfault>
I didn't; why do you need it?
<bbigras>
maybe the fast connection thing doesn't do the thing I think it does but when I update I generally update my desktop first. and then my laptop (but I ran the command on my desktop). I think that having the fast connection thing set for my laptop would make it so that files would be copied from my desktop instead of the internet.
<lovesegfault>
all that fast-conn does is it _doesn't_ set --substitute-on-destination when performing the `nix copy` for the system drv
<lovesegfault>
(c.f. `nix copy --help`)
<lovesegfault>
so, indeed, it's the behavior you're seeing
<lovesegfault>
which allows you to speed things up if the conn between two machines is faster than between them and your binary substituters
<lovesegfault>
IME it's rarely the case
<bbigras>
isn't my connection between my desktop and laptop faster than my internet connection. unless maybe my harddrives suck
<bbigras>
ssd*
<lovesegfault>
It depends. Your wifi can suck at upload
<lovesegfault>
mine, for example, does
<bbigras>
oh yeah
<lovesegfault>
my laptop, where I mostly do builds, sucks at uploading via wifi
<__monty__>
Ah the asymmetry of infrastructure mode strikes again : )
<lovesegfault>
so it's actually faster to just let my desktop fetch from the web
<bbigras>
yeah and with unlimited internet we don't care about how much we download
<bbigras>
and maybe while updating my laptop I might enjoy than my desktop has low disk usage while I'm using it
<aanderse>
when you crack open your PC, swap a sata slot, then 2 hours later you realize your entire network has fallen apart because changing a sata slot caused `enp2s0` to become `enp3s0`
<aanderse>
-_-
<bbigras>
wat
<joepie91>
~predictable interface names~
<aanderse>
the most painful part of that being the last 30 minutes when you were analyzing iptables rules, not noticing that your new device was enp3s0, not enp3s0
Fare has joined #nixos-chat
<aanderse>
yeah I'm still not sure why the called it "predictable"
<aanderse>
in any scenario I've ever actually experienced it is the exact opposite
<etu>
aanderse: They are more predictable than before
<sphalerite>
it's predictable in that it'll stay the same as long as you don't alter the hardware setup. Which is a lot better than without it, where interfaces can swap names randomly between boots!